STA Class 6: Nonlinear Relationships

• Extrapolation is attempting to use the model to predict a value that at lies outside the dataset.

• “Nonlinear” indicates that the linear assumption must have been violated when we attempted to crease a model

Interpretations

• Linear: Y = mx + b —> 1 unit increase in X causes a m unit increase in Y

• Log-linear: log(Y) = mx + b —> 1 unit increase in X causes a (100)m% increase in Y

• Linear-log: Y = mlog(x) + b —> 1% increase in X causes a (.01)m increase in Y

• Log-log: log(Y) = mlog(x) + b —> 1% increase in X causes a m% increase in Y